An Evaluation of Immersive Displays for Virtual Human Experiences Kyle Johnsen1, Benjamin Lok2 Department of Computer and Information Science and Engineering University of Florida
ABSTRACT This paper compares a large-screen display to a non-stereo head-mounted display (HMD) for a virtual human (VH) experience. As VH experiences are increasingly being applied to training, it is important to understand the effect of immersive displays on user interaction with VHs. Results are reported from a user study (n=27) of 10 minute human-VH interactions in a VH experience which allows medical students to practice communication skills with VH patients. Results showed that student self-ratings of empathy, a critical doctor-patient communication skill, were significantly higher in the HMD; however, when compared to observations of student behavior, students using the large-screen display were able to more accurately reflect on their use of empathy. More work is necessary to understand why the HMD inhibits students’ ability to self-reflect on their use of empathy. Keywords: virtual humans, embodied agents, display comparison, medical education, immersive virtual environments Index Terms: I.3.7 [Computer Graphics]: Three-Dimensional Graphics and Realism Virtual Reality. H.5.2 [Information Systems and Presentation]: User Interfaces – Evaluation 1
INTRODUCTION
Rapidly gaining in popularity, virtual human (VH) experiences are virtual reality systems that focus on providing immersive social experiences. VH experiences are being used for many applications including military leadership training [1], social phobia treatment[2], and health professions training [3-5]. In VH experiences, the user’s primary task is to interact with one or more VHs. We compare two very different immersive visual displays, a head-mounted display (HMD) and a large-screen display. The choice of these two displays was made because they 1) are commonly used in immersive virtual reality, 2) are widely available, 3) have similar infrastructure and monetary costs, and most importantly 4) enable close-up, life-size VHs. Results are reported from a user study, which evaluated the displays’ impact on user perception and behavior in the VH experience. Display systems may influence users based on their characteristics. Physical display characteristics (e.g. size, weight, wires, and tracking equipment) affect what the user can do in the experience. Intrinsic display characteristics (e.g. resolution, fieldof-view, brightness, head-tracking, and stereoscopy) affect what the user perceives. There are many examples in the general 1
[email protected] [email protected]
Figure 1. The user interacts with a close, life-size VH using either a large-screen display (top) or HMD (bottom)
virtual reality literature where task performance is impacted by display choice, e.g. [6-9]. In a VH experience, the impact tends to be behavioral or perceptual. Zanbaka et. al. point out that the smaller field-of-view of an HMD allows users to easily ignore a VH by turning their heads slightly away from the VH [10]. When the VH is embedded into the real-world on a large-screen display, they are part of the users’ larger field-of-view. Also, immersion has a powerful behavioral effect on user behavior in H-VH interactions. In [11], an immersive display amplified user anxiety when speaking to an audience of VHs. This work brings immersive displays research together with an important real-world application, using a VH experience for medical communication skills training. For conducting medical interviews, medical students must develop strong doctor-patient communication skills. Communication skills include the sequence and type of questions, listening, and expressing empathy. A VH experience was constructed that allows students to practice and be evaluated on these skills, called the Interpersonal Simulator (IPS) [4]. The experience has been used by hundreds of medical students, and has been validated for doctor-patient communication skills education [12]. The motivation behind this work is to understand if the choice of immersive visual display can impact students’ perception and/or behavior in the VH experience.
2
THE INTERPERSONAL SIMULATOR (IPS)
The IPS combines immersive displays, tracking, natural speech processing, and realistic VHs to produce an engaging VH experience. Three networked computers were used to run the IPS during the study. One ran speech recognition and understanding (Intel Pentium 4, 3.4 GHz), another performed rendering and animation (Intel Core 2, 2 GHz, NVidia GeForce 7900), and the last was used for the optical tracking system (Intel Core 2, 2 GHz). The frame rate for the IPS was consistently greater than 60 frames per second. 2.1 Immersive Displays The IPS can be switched between two immersive displays, a non-stereo head-mounted display (HMD), and a fish-tank projection display (FTPD). Note: Ideally, both of these displays would support stereo viewing; however, the FTPD does not support stereo at this time. Thus the HMD was only used in mono mode (the same image for both eyes). However, subjectively the difference between stereo and mono for the HMD is small for the interview task. The FTPD is a type of large-screen display; “fish-tank” because the displayed image is rendered from the perspective of the user as in [13]. To produce the large screen it uses a projector. The projector for the study was an NEC LT260. The resolution was set to 800x600 to match the native resolution of the HMD. In this way, the virtual world and the VH are brought to the user, rather than the user being brought to the VH as in the HMD. The user’s head position is tracked using a custom built optical tracker similar to the one described in [14]. The update rate for this tracker is 60 Hz. The tracking accuracy is approximately 1 cm for the examination room space. The jitter is less than 0.5 mm, and the precision is variable depending on marker size, but is less than 1 mm in practice. The end-to-end latency is less than 100ms. Features for this type of display relative to the HMD are: • The user is not encumbered by wires or the display • The field-of-view is on average larger than the HMD (60-100° diagonal) • The user can see their own body • The user can interact with real-world objects easily (e.g note taking) The HMD used was an eMagin Z800 3D Visor. By tracking the orientation and position of the HMD and coupling them with rendered viewing direction and position, the user is given the impression that they are in the virtual world, and can look in any direction. The Z800 is low-cost (